Estimating Bulk Geometrical Properties of Cellular Structures

TitleEstimating Bulk Geometrical Properties of Cellular Structures
Publication TypeThesis
Year of Publication2003
AuthorsPuddister, S.
Academic DepartmentDepartment of Systems Design Engineering
UniversityUniversity of Waterloo
CityWaterloo, Ontario, Canada
Thesis TypeM.A.Sc. Thesis
Abstract

Three algorithms for determining the bulk geometric properties of cellular structures from a stationary digital image are presented. The geometrical properties of interest include cell orientation and cell aspect ratio. With future test data expected to possess high luminance variability, poor contrast near element boundaries and irregular discolouration, this problem is poorly suited for the spatial techniques currently in the computer vision literature. Algorithms operating within the spatialfrequency domain based on Gabor filters, area moments (moments of inertia) and least squares ellipse fitting are used to determine the bulk geometrical properties of cellular structures in a digital image.